Among substance abusers, post-partum mothers offer an opportunity for identification, referral and treatment that is not matched by any other group. Drug screens administered to newborn babies are capable of leading to more certain identification of substance abuse among mothers than would be afforded even by random testing of mothers, something which is not currently possible. The period of hospital confinement and followup care provides an opportunity to establish a relationship that can be exploited for referral to drug treatment. A highly active referral process, combining psychological support with practical assistance, has achieved a success rate of over 50% in leading to acceptance of referral by a group of adolescent post-partum mothers. An equally active follow process has kept the majority in treatment. Successful application of this identification-referral-treatment process to post-partum mothers in general would extend the benefits of treatment to a population of substance abusers which would otherwise continue to be a risk not only to themselves but to their children present and future. The study here proposed would assess the effectiveness of the identification-referral-treatment process through a random experiment involving 600 identified substance abusing post-partum mothers. To allow the emergence of any significant interactions between the referral process and the type of treatment through which clients are referred, treatment will be provided under the two principal competing options: an outpateint treatment by trained substance abuse therapists and residential treatment involving a community of post-partum mothers and their children living and working together under the supervision of treatment professionals. The effectiveness of the treatment will be assessed through indepth interviews conducted by an independent organization experienced in assessing drug use, as well as general life adjustment. The interviews will be conducted prior to and two years following random assignment to the two treatment and one control group.